Función del preservativo
COSTO UNITARIO
It is important that users can express their subjective perceptions about music at every level, e.g., music artist, song, genre or compilation. This information is processable by the PMKB KMS. For instance, when a user associates a specific mood or occasion with a self created playlist. All this subjective knowledge can be used for further tasks, e.g., a music recommendation based on a specific mood.
On the other side, this knowledge can influence evaluation tasks to satisfy specific knowledge requests, e.g., when a user re-defines a description of a specific music genre or she/he pro- vides specific preferences regarding specific environmental contexts. In this case a user creates idiosyncratic genres or music contexts (see Section 2.4.2). Knowledge request processing is strongly dependent on the user and the context and probably delivers quite different results for the same query (see Section 2.4).
To enable such a knowledge processing in a multiple device environment, which is common nowadays, these KRs of personal associations and contexts have to be synchronised with the GKRM (see Subsection 3.2).
3.4
SUMMARY
The important features of the concept personal music knowledge base are given in the following short list:
• The knowledge representation model has to be able to represent music content and con- text, user profile and profiling data (especially personal associations and context, see Sub- section 3.3.6) in the same manner (see Subsection 3.1.1)
• The knowledge management system has to be able to handle the four important information processing capabilities (creatable, accessible, modifiable, removable), data synchronisation, access control, and knowledge exchange and reasoning (see Subsection 3.1.2)
• The PMKB has a global music knowledge base and local music knowledge bases (see Subsection 3.2)
• The PMKB has a strong concern on information federation, personalisation and timeliness/- data synchronisation (see Subsection 3.3)
KNOWLEDGE BASE
This chapter describes an exemplary, platform specific realisation of the concept personal music knowledge base. The PMKB is described as mainly computational and fully platform independent concept in the previous chapter. This chapter is aligned to the structure of the "abstract concept" chapter, to map the features and requirements of the abstract PMKB description directly onto the concrete concept. This instantiation is on a very high level platform specific, i.e., only main technologies, e.g, KRLs (vocabularies) and protocols are defined. A detailed explanation of novel designed or massively modified KR vocabularies is given in Section 4.1. This part is followed by a small summary of design decisions, that are made on this implementation level, regarding the PMKB KMS (see Section 4.2).
4.1
KNOWLEDGE REPRESENTATION
A conclusion of Section 3.1.1 is that an ontological KRM is a good choice to satisfy the needs of a KRM for the PMKB. As introduced in Section 2.2.4, Semantic Web KRLs and ontologies are an appropriate basis to satisfy the needs of such a KRM. One can
• model universals (classes and properties) and particulars (individuals) with them in the same manner,
• separate domain specific descriptions into separate ontologies (which can also have a dif- ferent level of detail),
• apply inferencing rules on them for reasoning and • name resource in a uniform way by utilising URIs.
To model the main domains that can be part of a KRM for the PMKB, one should try reusing existing Semantic Web ontologies as much as possible. Firstly, the Music Ontology framework is an appropriate entry point for music content and context data (see Section 2.3.2). Secondly, FOAF Vocabulary is a good foundation for describing user profiles (see 2.4.1).
However, since this knowledge can be federated from several, selectable information services, we also need an ontological KR to describe them (see Subsection 4.1.1). We require further
ontological KRs to be able to record user behaviour, describe recommendations, personal asso- ciations and environmental context (see Subsections 4.1.2.4, 4.1.3, and 4.1.2.3). Moreover, some additions for modelling users are useful and necessary (see Subsection 4.1.4.2). Finally, a taxon- omy for media types closes a further gap that was discovered during the implementation of the PMKB concept (see Subsection 4.1.5).
All these ontological KRs are expressed in several new Semantic Web ontologies, which can be included into the PMKB KRM. They are presented in the following subsections1.
existing vocabulary modified vocabulary new vocabulary Music Ontology
Info Service Ontology
Weighting Ontology Property Reification Vocabulary
Web Ontology Language Recommendation Ontology Play Back Ontology
Association Ontology
Similarity Ontology
Ordered List Ontology
Counter Ontology
Event Ontology Review Vocabulary
Cognitive Characteristics Ontology Bibliographic Ontology
Functional Requirements for Bibliographic Records Vocabulary
DCMI Metadata Terms
Friend of a Friend Vocabulary
Statistical Core Vocabulary
Resource Description Framework
OWL Time Ontology Resource Description Framework Schema
XML Schema
Figure 4.1: An overview of applied ontologies and vocabularies (see [Gän10f])
Figure 4.1 shows an overview of applied ontologies and vocabularies in the PMKB KRM. On the one side, one can see the reutilisation of existing Semantic Web ontologies in new ones. On the other side, this graphic illustrates the application of the new vocabularies in other new ones, too. The majority2of the represented (re-)utilised existing KRLs and ontologies are already introduced in Chapter 2. A link from one vocabulary to another one was created, when at least one sub class, sub property, domain or range relation to another vocabulary exists. Some existing Se- mantic Web ontologies were also modified during the development process for the PMKB KRM. 1KRM examples are always serialised in N3 (see Section 2.2.2). Graphical illustrations in this section are mainly composed with the help of TopBraid Composer (see Section 2.2.4.1).
2Please have a look at [Gän10f] for references to those KRLs and vocabularies which are not separately introduced in this thesis.
The Property Reification Vocabulary is explicitly outlined in Subsection 4.1.4.3, due to its notewor- thy modifications. An overview of new created or initially published Semantic Web ontologies, vocabularies and taxonomies in this context is given in [Gän10g].